Chronic administration of cocaine to animals depletes brain stores of serotonin, and serotonin receptors may be upregulated through the mechanism of denervation sensitivity. Treatment with L-tryptophan, the amino acid precursor of the neurotransmitter serotonin, will replenish serotonin stores and may hasten the down-regulation of serotonin receptors. We hypothesize that cocaine withdrawal symptoms may be produced, in part, by serotonin depletion. This is the first controlled clinical study to evaluate the efficacy of L-tryptophan (5 gm/day) in reducing cocaine withdrawal symptoms and post-treatment relapse to cocaine use. During this three year study, 225 subjects will be recruited from inpatients undergoing treatment for cocaine abuse at Merritt Peralta Chemical Dependency Recovery Hospital, a 40 bed, free-standing, not-for-profit, acute care hospital in Oakland, California. Using a parallel group design, subjects will be randomly assigned to one of three groups. Groups 1 and 2 will be double-blind (Group 1 will receive L-tryptophan and Group 2 will received placebo). Subjects assigned to Group 3 will receive open label L-tryptophan. Group 3 is a comparison group to measure the effects of subjects knowing that they are participants in placebo controlled research. Treatment with L-tryptophan (or placebo) lasts 28 days while the subject is hospitalized. Follow-up assessment will be done 30 and 60 days post-discharge. Biological, subjective and behavioral data will be collected. Biological measures include serum levels of cocaine, benzoylecgonine, prolactin, and urine testing for drugs of abuse. Subjective measures include the MMPI, POMS, the SCL-90, cocaine craving scale. Behavioral outcome measures include ASI, length of stay in treatment, post-discharge drug use, number of days of use, and time to relapse. Data analysis compares rates of symptom improvement between L-tryptophan and placebo treated subjects by computing regression slopes for each subject's scale scores on the measures, computing the mean slope for each group and testing the difference between mean slopes with analysis of variance. Differences between Groups 1 and 2 measure the efficacy of L-tryptophan; differences between Groups 1 and 3 measure the pharmacological effect of being in a controlled group. Biological and psychological predictors of successful treatment outcome will also be studied.